使用drawtext过滤器后的ffmpeg串联

6

我对ffmpeg还比较新,但是在这个问题上搜索了几天后,我已经完全遇到了瓶颈。任何帮助都将不胜感激。

我的使用情况:我们的客户想要为多个地区上传视频。每个视频都将具有相同的格式,即1920x1080,mp4。对于每个地区,他们想要在视频末尾添加一个不同的图像,持续几秒钟。此图像包含其徽标、一些附加信息和一个可变代码。他们将在上传的视频旁输入此代码。该图像保持不变,因此已经存在于服务器上。 因此,我有一个输入视频、一个图像视频和一个小代码。我需要将此代码添加到图像视频中(在预定义位置),然后我需要将生成的视频添加到输入视频的末尾。完成后,我只需要以1920x1080和1024x576输出视频。

我尝试了几种方法,但连接步骤总是失败,操作过的视频也无法正常播放。

尝试1

在第一次尝试中,我使用ffmpeg从图像创建视频,并在指定区域添加文本。

ffmpeg -y -f lavfi -i image.png -r 30 -t 10 -pix_fmt yuv420p -map 0:v -vf drawtext="fontfile=HelveticaNeue.dfont: text='GLNS/TEST/1234b': fontcolor=black: fontsize=20: box=1: boxcolor=white: boxborderw=7: x=179: y=805" imageVideo.mp4

这个命令可以创建一个正确大小、持续时间为10秒的.mp4视频,并在正确位置添加文本“GLNS/TEST/1234b”。
接下来,我使用以下命令将两个视频连接起来。两个视频具有相同的分辨率和编解码器。
ffmpeg -f concat -safe 0 -i config.txt -vf scale=1920:1080 outputHD.mp4 -vf scale=1024:576 outputSD.mp4

config.txt 包含以下内容:

file my_input_file.mp4
file ImageVideo.mp4

这个连接适用于常规视频。然而,当我将其与ImageVideo.mp4(由第一个命令创建的文件)一起使用时,会出现以下错误日志:

[mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f86dc924600] Auto-inserting h264_mp4toannexb bitstream filtereed=0.509x    
[aac @ 0x7f86dc019e00] Number of bands (31) exceeds limit (5).
Error while decoding stream #0:1: Invalid data found when processing input
[aac @ 0x7f86dc019e00] Number of bands (27) exceeds limit (8).
Error while decoding stream #0:1: Invalid data found when processing input
[h264 @ 0x7f86dd857200] Error splitting the input into NAL units.
[h264 @ 0x7f86dd829400] Invalid NAL unit size.
[h264 @ 0x7f86dd829400] Error splitting the input into NAL units.
[aac @ 0x7f86dc019e00] Number of bands (10) exceeds limit (1).
Error while decoding stream #0:1: Invalid data found when processing input
[h264 @ 0x7f86dd816800] Invalid NAL unit size.
[h264 @ 0x7f86dd816800] Error splitting the input into NAL units.
[aac @ 0x7f86dc019e00] Number of bands (24) exceeds limit (1).
Error while decoding stream #0:1: Invalid data found when processing input

#this goes on for a few hundred lines

输出结果与输入视频完全相同,但不包含最后期望的图像视频。

尝试2

由于上述尝试没有成功,我尝试了将我们的设计师使用Adobe After Effects制作的图像视频连接起来。这个视频也使用H264编解码保存为.mp4格式。如果我将输入视频和此视频连接起来,则可以得到正确的结果。然而,一旦我在指定区域添加代码并使用此命令:

ffmpeg -i new_image_video.mp4 -vf drawtext="fontfile=HelveticaNeue.dfont: text='GLNS/TEST/1234b': fontcolor=black: fontsize=20: box=1: boxcolor=white: boxborderw=7: x=179: y=805" -c:v libx264 imageVideo.mp4

我收到了这个错误信息:

[mov,mp4,m4a,3gp,3g2,mj2 @ 0x7ff94c800000] Auto-inserting h264_mp4toannexb bitstream filter97x    
[h264 @ 0x7ff94b053800] top block unavailable for requested intra mode -1
[h264 @ 0x7ff94b053800] error while decoding MB 0 0, bytestream 49526
[h264 @ 0x7ff94b053e00] number of reference frames (1+3) exceeds max (3; probably corrupt input), discarding one
[h264 @ 0x7ff94b053e00] chroma_log2_weight_denom 28 is out of range
[h264 @ 0x7ff94b053e00] illegal long ref in memory management control operation 2
[h264 @ 0x7ff94b053e00] cabac_init_idc 32 overflow
[h264 @ 0x7ff94b053e00] decode_slice_header error
[h264 @ 0x7ff94b053e00] no frame!
[h264 @ 0x7ff94b053800] concealing 8160 DC, 8160 AC, 8160 MV errors in I frame
[h264 @ 0x7ff94b072a00] reference overflow 22 > 15 or 0 > 15
[h264 @ 0x7ff94b072a00] decode_slice_header error
[h264 @ 0x7ff94b072a00] no frame!
[h264 @ 0x7ff94b01a400] illegal modification_of_pic_nums_idc 20
[h264 @ 0x7ff94b01a400] decode_slice_header error
[h264 @ 0x7ff94b01a400] no frame!
[h264 @ 0x7ff94b01aa00] illegal modification_of_pic_nums_idc 20
[h264 @ 0x7ff94b01aa00] decode_slice_header error
[h264 @ 0x7ff94b01aa00] no frame!
Error while decoding stream #0:0: Invalid data found when processing input
[h264 @ 0x7ff94b053800] deblocking_filter_idc 8 out of range
[h264 @ 0x7ff94b053800] decode_slice_header error
[h264 @ 0x7ff94b053800] no frame!
Error while decoding stream #0:0: Invalid data found when processing input
[h264 @ 0x7ff94b053e00] illegal memory management control operation 8
[h264 @ 0x7ff94b053e00] co located POCs unavailable
[h264 @ 0x7ff94b053e00] error while decoding MB 2 0, bytestream -35
[h264 @ 0x7ff94b053e00] concealing 8160 DC, 8160 AC, 8160 MV errors in B frame
[h264 @ 0x7ff94b072a00] number of reference frames (1+3) exceeds max (3; probably corrupt input), discarding one

# this goes on for a while...

[h264 @ 0x7ff94b01a400] concealing 4962 DC, 4962 AC, 4962 MV errors in B frame
Error while decoding stream #0:0: Invalid data found when processing input
frame= 2553 fps= 17 q=-1.0 Lsize=   26995kB time=00:01:42.16 bitrate=2164.6kbits/s dup=0 drop=60 speed=0.697x    
video:25258kB audio:1661kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.285236%
[libx264 @ 0x7ff94b810400] frame I:35    Avg QP:17.45  size: 55070
[libx264 @ 0x7ff94b810400] frame P:711   Avg QP:19.73  size: 18712
[libx264 @ 0x7ff94b810400] frame B:1807  Avg QP:21.53  size:  5884
[libx264 @ 0x7ff94b810400] consecutive B-frames:  3.4%  5.0%  4.9% 86.6%
[libx264 @ 0x7ff94b810400] mb I  I16..4: 38.2% 49.3% 12.5%
[libx264 @ 0x7ff94b810400] mb P  I16..4: 12.4% 14.0%  1.0%  P16..4: 29.6%  4.8%  1.9%  0.0%  0.0%    skip:36.2%
[libx264 @ 0x7ff94b810400] mb B  I16..4:  1.5%  1.2%  0.1%  B16..8: 27.3%  1.6%  0.1%  direct: 1.8%  skip:66.4%  L0:45.8% L1:51.4% BI: 2.8%
[libx264 @ 0x7ff94b810400] 8x8 transform intra:49.5% inter:85.4%
[libx264 @ 0x7ff94b810400] coded y,uvDC,uvAC intra: 21.2% 22.3% 2.5% inter: 4.6% 7.0% 0.0%
[libx264 @ 0x7ff94b810400] i16 v,h,dc,p: 23% 26% 10% 41%
[libx264 @ 0x7ff94b810400] i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 31% 19% 35%  3%  3%  3%  3%  3%  2%
[libx264 @ 0x7ff94b810400] i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 31% 20% 16%  5%  7%  6%  5%  5%  4%
[libx264 @ 0x7ff94b810400] i8c dc,h,v,p: 67% 16% 15%  2%
[libx264 @ 0x7ff94b810400] Weighted P-Frames: Y:7.3% UV:4.2%
[libx264 @ 0x7ff94b810400] ref P L0: 66.3%  8.7% 17.9%  7.0%  0.1%
[libx264 @ 0x7ff94b810400] ref B L0: 88.2% 10.1%  1.7%
[libx264 @ 0x7ff94b810400] ref B L1: 94.9%  5.1%
[libx264 @ 0x7ff94b810400] kb/s:2026.12
[aac @ 0x7ff94b072400] Qavg: 635.626

生成的输出与输入视频完全相同,但不包含所需的图像视频在结尾处。

我注意到一件事:当我在Mac上检查视频文件(获取信息)时,它们总是包含这些行在“更多信息”中:

Dimensions: 1920 x 1080
Codecs: H.264, AAC
Color profile: HD(1-1-1)
Duration: 01:42
Audio channels: 2
Last opened: Today 11:02

然而,通过drawtext过滤器的视频具有以下特点:
Dimensions: 1920 x 1080
Codecs: AAC, H.264
Duration: 00:10
Audio channels: 2
Last opened: Today 11:07

如您所见,这里没有颜色配置文件,并且编解码器的位置已经改变。我认为这与我的问题有关,但是我似乎找不到解决方法。
PS:该应用将在php环境(Symfony)中运行。 我注意到Symfony绑定的ffmpeg中并未提供concat命令,因此我使用常规终端命令来执行此操作,将使用php执行这些命令。
编辑 尝试3 根据同事的建议,我尝试将视频转换为.avi格式,然后重新转换为.mp4格式,以便去除由drawtext滤镜包含的任何损坏或额外信息。但是这会导致完全不同的错误。
[mov,mp4,m4a,3gp,3g2,mj2 @ 0x7f812413da00] Auto-inserting h264_mp4toannexb bitstream filtereed=0.516x    
[concat @ 0x7f8124009a00] DTS 1569260 < 2551000 out of order
[h264 @ 0x7f8124846800] left block unavailable for requested intra4x4 mode -1
[h264 @ 0x7f8124846800] error while decoding MB 0 0, bytestream 47919
[h264 @ 0x7f8124846800] concealing 8160 DC, 8160 AC, 8160 MV errors in I frame
[aac @ 0x7f8125809a00] Queue input is backward in time
[aac @ 0x7f8125815a00] Queue input is backward in time
[h264 @ 0x7f8124846e00] number of reference frames (1+3) exceeds max (3; probably corrupt input), discarding one
[h264 @ 0x7f8124846e00] chroma_log2_weight_denom 26 is out of range
[h264 @ 0x7f8124846e00] deblocking_filter_idc 32 out of range
[h264 @ 0x7f8124846e00] decode_slice_header error
[h264 @ 0x7f8124846e00] no frame!
[mp4 @ 0x7f8124802200] Non-monotonous DTS in output stream 0:1; previous: 4902912, current: 4505491; changing to 4902913. This may result in incorrect timestamps in the output file.
[mp4 @ 0x7f8125813000] Non-monotonous DTS in output stream 1:1; previous: 4902912, current: 4505491; changing to 4902913. This may result in incorrect timestamps in the output file.
[h264 @ 0x7f8124803400] reference overflow 20 > 15 or 0 > 15
[h264 @ 0x7f8124803400] decode_slice_header error
[h264 @ 0x7f8124803400] no frame!
[mp4 @ 0x7f8124802200] Non-monotonous DTS in output stream 0:1; previous: 4902913, current: 4506515; changing to 4902914. This may result in incorrect timestamps in the output file.
[mp4 @ 0x7f8125813000] Non-monotonous DTS in output stream 1:1; previous: 4902913, current: 4506515; changing to 4902914. This may result in incorrect timestamps in the output file.
[mp4 @ 0x7f8124802200] Non-monotonous DTS in output stream 0:1; previous: 4902914, current: 4507539; changing to 4902915. This may result in incorrect timestamps in the output file.
[mp4 @ 0x7f8125813000] Non-monotonous DTS in output stream 1:1; previous: 4902914, current: 4507539; changing to 4902915. This may result in incorrect timestamps in the output file.

# Again, this continues for quite a while.
1个回答

3
请使用 concat 过滤器,并在一个命令中完成。
ffmpeg -i main.mp4 -loop 1 -framerate 30 -t 10 -i image.png 
       -filter_complex "[1]drawtext=fontfile=HelveticaNeue.dfont:text='GLNS/TEST/1234b':
                           fontcolor=black:fontsize=20:box=1:boxcolor=white:boxborderw=7:
                           x=179:y=805[it];
                        [0:v][it]concat,split[hd][sd]"
      -map "[hd]" -map 0:a combined1080p.mp4
      -map "[sd]" -map 0:a -s 1024x576 combined576p.mp4

我假设您的主视频和图像已经是1920x1080大小。

3
对于我们中的新手,为什么这个方法有效,而另一个失败了? - vrwim
非常感谢,太棒了,完美解决了。不过我得删除 drawtext= 后面的引号,那是个笔误。像 vrwim 问的那样,您能否解释一下为什么这样做有效以及出了什么问题? - Sven Hoskens
1
我无法确定地调试您的命令,但是concat demuxer需要匹配的流顺序和匹配的流属性,例如编解码器、分辨率、像素格式(虽然不需要色彩空间),但还需要诸如时间基准、0起始PTS等属性,这些属性不能由应用程序(例如AE)允许用户修改或新的FFmpeg用户不知道如何指定。当您拼接从同一源生成的段时,最好使用concat demuxer。 - Gyan

网页内容由stack overflow 提供, 点击上面的
可以查看英文原文,
原文链接